Question 70
At an AI model development checkpoint, an AI project leader faces this situation: A model meets offline accuracy requirements but its production container cannot scale to the required request volume. The chosen approach must remain reproducible and satisfy release criteria. Which use case most closely matches this proposed approach?
Explanation
Correct Answer
In an AI model development checkpoint, this is the matching use case: Use serving-environment acceptance tests before model cutover. The reason is: Operational readiness includes packaging, capacity, latency, observability and security of the model service. In the reported scenario, a model meets offline accuracy requirements but its production container cannot scale to the required request volume. The appropriate action is: Test the production-like serving environment under representative peak load and failure conditions.
Incorrect Answers
Preprocessing must distinguish absent values from meaningful numerical zero and preserve business semantics. The appropriate correction for the reported case is: Test the production-like serving environment under representative peak load and failure conditions.
Data readiness decisions require documented metrics, quality exceptions and accountable approval. Fixing that different mechanism would not address the current root cause: operational readiness includes packaging, capacity, latency, observability and security of the model service.
Algorithm selection must reflect available labels and whether the objective is prediction, grouping or another task. Its verification is distinct from the reported condition: A model meets offline accuracy requirements but its production container cannot scale to the required request volume. The chosen approach must remain reproducible and satisfy release criteria.